Year and Industry dummies with GEE

[Richard Krombach]

Hello everyone,

I have an urgent problem. I am investigating the implications of CEO´s political ideology and Corporate Strategic Change using Panel Data. I use a GEE to do so, which is common practice for these type of analysis and it is also approved by my supervisor. The problem I have is that as soon as I include year and industry dummies in my GEE regression, no Wald Chi2 is calculated. I use the following code:

xtset co_per_rol fyear
xtgee leading_strategic_change ceo_conservatism industry_control pre_entry ceo_tenure firm_age performance firm_size employees ceo_duality ceo_age incentive_compensation democratic_president  liberal_state i.fyear i.sic_industry, vce(robust)

The results that I get are attatched as a PDF file Results.pdf .

Could any of you please help me and tell me if either my model or the code I am using is wrong?

Thank you very much in advance,

Richard

[Richard Krombach]

Maybe an additional hint: It seems to be a problem with the industry dummies: If create the individual dummies for each industry and only include the dummies for the industries 7-55 everything is working fine. However, if I only use the dummies for industries 1-6 I get the same problem. So there shouldnt be a problem with degrees of freedom, but some other problem. Thank you again

[Joro Kolev]

What makes the problem urgent, Richard? If you do not figure out as soon as possible how CEO political ideology affects Corporate strategic change, the great democracy of the US will cave in (in no time), and we in the rest of the world would not have a stellar democratic exemplar to emulate ?

Can you show summary statistics for your variables? The most important are the summary statistics for the dependent variable and your main regressor.

Most probably the missing chi2 is some numerical problem.

[Richard Krombach]

What makes the problem urgent, Richard? If you do not figure out as soon as possible how CEO political ideology affects Corporate strategic change, the great democracy of the US will cave in (in no time), and we in the rest of the world would not have a stellar democratic exemplar to emulate ?

Can you show summary statistics for your variables? The most important are the summary statistics for the dependent variable and your main regressor.

Most probably the missing chi2 is some numerical problem.

. sum leading_strategic_change

Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
leading_st~e | 11,476 -1.304094 2.136134 -2.8057 18.12317

. sum ceo_conservatism

Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
ceo_conser~m | 12,536 .6417399 .3800217 0 1

These are the summary statistics on strategic change (DV) and Political ideology.
. tabulate sic_industry

sic_industr |
y | Freq. Percent Cum.
------------+-----------------------------------
1 | 41 0.33 0.33
2 | 11 0.09 0.41
10 | 83 0.66 1.08
12 | 58 0.46 1.54
13 | 557 4.44 5.98
14 | 38 0.30 6.29
15 | 111 0.89 7.17
16 | 81 0.65 7.82
17 | 47 0.37 8.19
20 | 394 3.14 11.34
21 | 39 0.31 11.65
22 | 49 0.39 12.04
23 | 109 0.87 12.91
24 | 83 0.66 13.57
25 | 158 1.26 14.83
26 | 206 1.64 16.47
27 | 60 0.48 16.95
28 | 1,116 8.90 25.85
29 | 162 1.29 27.15
30 | 113 0.90 28.05
31 | 36 0.29 28.33
32 | 48 0.38 28.72
33 | 184 1.47 30.19
34 | 238 1.90 32.08
35 | 775 6.18 38.27
36 | 718 5.73 43.99
37 | 462 3.69 47.68
38 | 727 5.80 53.48
39 | 102 0.81 54.29
40 | 91 0.73 55.02
41 | 12 0.10 55.11
42 | 133 1.06 56.17
44 | 81 0.65 56.82
45 | 159 1.27 58.09
47 | 55 0.44 58.53
48 | 426 3.40 61.93
49 | 1,114 8.89 70.81
50 | 223 1.78 72.59
51 | 190 1.52 74.11
52 | 56 0.45 74.55
53 | 143 1.14 75.69
54 | 69 0.55 76.24
55 | 77 0.61 76.86
56 | 184 1.47 78.33
57 | 65 0.52 78.84
58 | 228 1.82 80.66
59 | 209 1.67 82.33
70 | 43 0.34 82.67
72 | 72 0.57 83.25
73 | 1,221 9.74 92.99
75 | 43 0.34 93.33
78 | 19 0.15 93.48
79 | 136 1.08 94.57
80 | 281 2.24 96.81
82 | 124 0.99 97.80
83 | 13 0.10 97.90
87 | 216 1.72 99.63
99 | 47 0.37 100.00
------------+-----------------------------------
Total | 12,536 100.00


. codebook sic_industry

-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
sic_industry (unlabeled)
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Type: Numeric (float)

Range: [1,99] Units: 1
Unique values: 58 Missing .: 0/12,536

Mean: 43.316
Std. dev.: 19.4142

Percentiles: 10% 25% 50% 75% 90%
20 28 38 53 73


This is the basis of the industry dummy that I want to create. As you can see I have 58 different industries, however in the GEE only not all are included for some reason (e.g. sic_industry==1)